Mosaic processing - can't get backgrounds even

[cmassa]

I finally got my 6 panel mosaic of the Pelican and North America Nebula to assemble. I was having problems with star alignment and finally got it to work. But I still just cannot get the background levels even close on a couple of the panels. They are much lighter and I finally just forced it and clipped it a lot during final processing with histogram transformation. Here is my first go at processing it:


The panels on the right side of the image are much lighter which also causes the some of the large stars to look strange.

Any hints? I posted this question before and got some good answers but have not been able to apply them effectively. I have tried Linear Fit but I must be using it incorrectly because it usually makes the image look terrible, if there is any image left at all.

Thanks for any advice,
Chris

[vpcirc]

I think it's better to apply DBE etc and stretch them as equally as possible first. There's a great video showing how to use select points to remove any gradients on the PI resources page.

[cmassa]

I have been using DBE before trying to assemble the mosaic. The problem I have is that some of the panels are lighter than others, which DBE doesn't correct because its not really a gradient. I need to decrease the overall brightness of the offending panels, or increase the dark ones. The gradients within each panel are gone but they vary compared to each other. Maybe I'm not explaining it correctly. When combined into the mosaic, it is easy to see the differences. If you need I can try to post the unprocessed mosaic with just an STF applied where it is more obvious.

Also, I thought it was better to assemble the mosaic when still at the linear stage. If that is not true I can try processing all of the panels and then assembling them. Will GradientMergeMosaic work with non-linear images?

Thanks,
Chris

[RickS]

Did you turn on Frame Adaptation when you used StarAlignment?  It applies a linear fit function to each target image to match it to the reference image.

Cheers,
Rick.

[cmassa]

Yes I do have frame adaptation enabled. Does that do something to the brightness of the panels?

Chris

[RickS]

Yes I do have frame adaptation enabled. Does that do something to the brightness of the panels?

Chris

Yes, it should try to match them - have a look at the tooltip.  I'm playing with a 2x3 mosaic at present and Frame Adaption is doing a good job although I did have problems until I cleaned up a few of my panels (by removing a few low quality subs with high background levels and picking reference frames more carefully.)

Cheers,
Rick.

[cmarcus]

https://www.youtube.com/watch?v=H2r3ZYqCaP8

this procedure any help?

I've used it for this:
https://www.astrobin.com/136637/
and
https://www.astrobin.com/136805/

[cmassa]

Yes, that is the procedure I use to create mosaics but it just starts with the actually assembly of the mosaic, not how to best preprocess the images when the overall lightness doesn't match. I guess I could wait until next summer to take new ones but I was trying to get this set to work better.

Chris

[pfile]

although it's perhaps "cheating", i have in the past looked at the mean value of the backgrounds of the images, then applied pixelmath expressions to brighten/darken the offending images by the ratio of the background means of the target images and reference image.

rob

edit: in other words, K: $T*(reference/target) or $T*(target/reference) depending on whether or not you need to brighten or darken the target image.

[vpcirc]

On my latest, I eyeballed the histograms of all 4 panels to get them close as possible, then a light DBE and finally a gradient exterminator in the forbidden program. Worked wonderful